Doc store batching (#3)

* move table definition into separate file

* Add table structs and traits

* start replacing explicit table access with tables struct

* more table use

* try to hide db use a bit

* more tables use

* add read and modify fns everywhere

* First impl of batching

* Eliminate one out of 2 places where we need to store a table in an iterator

* shut up clippy!

* introduce ReplicaInfo

* pass in store to replica and replica_if_sync

WIP

* try to avoid using the actual replica

often you just need to modify the info or the state

* remove close

it seems to do nothing.

* simplify close callback

* remove the close callbacks entirely

* Add more trait impls

* separate loading info and creating the actual replica

* make replica a transient thing

* &mut T or T, your choice!

* Last step in making StoreInstance and Replica ephemeral (or at least optionally ephemeral)

* all tests passing

We can eliminate AlreadyOpen because rust will make sure we open only 1 replica now!

* get rid of the arc

* WIP

* back to .modify

due to Voultapher/self_cell#14

* don't import derive_more

* Rename the table access methods

modify(f) to modify
readonly for ro view of mutable table
snapshot for a real readonly snapshot

* fix test_migration_001_populate_latest_table test

you need to explicitly call flush now before copying the disk image!

---------

Co-authored-by: Ruediger Klaehn <[email protected]>

Author: ppodolsky

Cargo.lock

@@ -46,6 +46,7 @@ tokio-util = { version = "0.7", optional = true, features = ["codec", "io-util",
 tokio-stream = { version = "0.1", optional = true, features = ["sync"]}
 quinn = { version = "0.10", optional = true }
 futures = { version = "0.3", optional = true }
+self_cell = "1.0.3"
 
 [dev-dependencies]
 iroh-test = { path = "../iroh-test" }